Class::XPath 1.0 has been released. It's basically a straight port of the OO-tree XPath matcher I was talking about a few weeks ago. So far I've used it with a simple test class, HTML::TreeBuilder and in the larger application where the code was originally developed.
The next step is to work on supporting more XPath syntax, which probably means writing a real parser. I'm particularly interested in adding attribute and content matching support which I think I can use in my application. I'm tempted to go with Parse::RecDescent but I know what that will do to performance. Maybe I should check out some of the simpler Yacc derivatives on CPAN...
I've also been thinking about the possiblity of supporting some kind of tree analysis to speed up query matching. Right now a query like '//foo' has to hit every node in the tree to see if it has any children named 'foo'. A simple cache mapping names to lists of nodes could speed that up quite a bit, at the expense of building the cache the first time. The problem is that the tree might change between building the cache and query time, and Class::XPath isn't in a very good place to notice. I suppose it's more of an interface issue than anything else.
-sam
